Why use a pentode?

[ilimzn]

I have some measurements that disagree.

Well, I measured a bunch of various ones recently on a curve tracer and although they are of course all similar overall, and within what one would normally expect from tube variances, the differences are IMHO enough not to be too certain about how well it does with H3 as a distortion cancelling driver for a SE triode (or triode strapped tetrode/pentode). This my main worry. One exception I found: ECC801S, the premium version. Tight tolerances and no distinguishable difference between sections...
Otherwise, as a small signal tube (where you don't try to get the maximum possible linear swing out of it) it's very useful.

Someone mentioned the GU50. An ECC81/12AT7 is a good driver for a triode strapped GU50 although a bit short in gain if one feeds such a simple two-stage amop with a standard line-level signal, although a CDP whould do nicely. And with some attention it can perform distortion cancellation.

The GU50 itself is a great tube, even better considering it's price. There are couple of interesting features - fG1, G2, G3 are all brought out to pins, giving a huge variety of things one can do with it (BTW strapping G2 and G3 to the anode can get you a nice extension of maximum anode voltage as otherwise you can get flashover between G2 and G3). Then, it has aligned grids, which signifficantly reduces G2 current WRT anode current, down to BPT levels - always nice in pentode mode. It's anode is also treated WRT the original LS50 it was modeled from and it can withstand a lot of abuse, ditto G2 and G1 are not exactly flimsy. And it's cheap - in fact the socket and carrier are often more expensive :|

 

[TonyTecson]

would anyone know if the GU50 is still in current production?

 

[Junm]

Chinese FU-50 are in current production...dunno about GU-50....mine are from 1970's.

 

[nigel pearson]

This tread started with input pentodes . I seem to remember EF 184 is much like an ECC 81 when in triode . That offers the interesting possibility to use a bit of its pentode curve to cancel some distortion if the Gu/Fu ( LS ) 50 is in triode .

Anyone know the EF184 ?

When reading my inherited TV magazines I found many tubes I thought to be audio devices in the TV section . PCL 86 if memory in correct ? EF184 also ?

 

[Tubelab_com]

Anyone know the EF184 ?

The EF184 is the same as the US 6EJ7, it was designed for the IF amp in TV sets. There is a related remote cutoff version the 6EH7/EF183. These can be found for $1 or less, but are generally not good for audio except in LTP and variable gain amps.

The 6EJ7 has been used in some DIY audio projects on this forum. The first that comes to mind is Pete Milletts latest P-P driver board. I have used it too because I have plenty of them.

I was working in TV repair when this tube started appearing in US TV sets. Most of the early tubes were of European origin that had been relabled by US tube and TV vendors. Sometime in the early 70's production started in the US.

The 6EJ7/EF184 had the highest Gm of any IF amp of the day, and some TV sets were on the verge of instability. Often you had to try two or three new tubes to fix a set that oscillated on weak signals. Some sets would be OK with US made tubes but not Philips tubes, and vice versa.

 

[ilimzn]

EF184 in triode is quite linear and has a much higher gm than an ECC81, and somewhat loer mu, hence also lower Rp. All quite desirable if you want to drive a big output tube with it.

PCL86 = ECL86 with different heater, roughly half of an ECC83 (12AX7) and dissipation limited EL84 in the same bulb. Have a look also at PCL85/805.

 

[nigel pearson]

If this thread continues I might copy some old TV circuits to show what we take to be audio tubes in what I am told is their true home . My fascination with PCL ECL( UCL ? ) 82/4/6 came from Dansette record players . Dansette if upgraded becomes a SE amp . I am proud to say I got 8 watts with similar sensitivity and 1 % THD from two tubes . If I had a Decca Deram pick up I probably could drive it directly . The Dansette motor even powered the heater . I suspect it wasn't ECL for that reason ? The UK 240 V makes a nice HT ( >320 DC ) . The bad bit is the pick up is more or less live . Hey Ho it was fine . I never remember anyone getting hurt . The sound was very reasonable . PYE Black Box was the posh version . My old boss lent one for the film Gangster No1 . Also some hi fi . The Dansette made more appearances . Not the best film I ever saw .

7199 I think is a better version of a TV tube ? Now that's a nice input pentode with a very handy triode for a direct coupled phase splitter . Radford in the UK had similar. They used a sort of LTP . Pentode side input . Said to be a bit special ( cancellation ? )

 

[Johan Potgieter]

As an aside regarding the GU50,

Looking up the data, I think it is the first time I saw a derating Pa graph with ambient temperature!

This very often overlooked characteristic is universal, not just for the GU50.

(As a further aside: I dumped ECC81 in favour of ECC88. Although with lower gain, I found the latter more linear. Comments?)

 

[ilimzn]

As a further aside: I dumped ECC81 in favour of ECC88. Although with lower gain, I found the latter more linear. Comments?

Do you mean the ECC88 being more linear than the ECC81 - which is no surprise since ECC81 was deliberately constructed with variable grid pitch, i.e. variable mu, while ECC88 is intended to be linear - it's non-linear cousin being the ECC189. Or do you mean in combination with a GU50, used as a driver for it?

 

[Johan Potgieter]

Ilimzn,

Serious? Thats news - I stand to be informed! But then how could the '81 be considered for audio work, particularly large signal?

No, I did mean ECC88 for audio in general. But then so is the '82 intended, and that one also has quite a bit of distortion. Next time I have a shot '81, I will have a look inside under the magnifier.

 

[Miles Prower]

Ilimzn,

Serious? Thats news - I stand to be informed! But then how could the '81 be considered for audio work, particularly large signal?

I don't think so. The spec sheet makes no mention of this, and the plate characteristics don't look like it either. There are some triodes made for TV front ends (the 6GK5) that do mention this specifically ("gain controlled triode") and some (the 6BQ7) with plate characteristics that look suspiciously like a variable-u characteristic is an undocumented feature. Never had the occasion to use the ECC81/12AT7, but they are seen frequently as driver devices due to the higher current capability than types like the 12AX7. Decent, low distortion loadlines aren't hard to find.

Although ECC81s (and the similar 12AT7) mention usage primarily as RF devices (oscillator/mixers, VHF amps) with audio being secondary.

No, I did mean ECC88 for audio in general. But then so is the '82 intended, and that one also has quite a bit of distortion. Next time I have a shot '81, I will have a look inside under the magnifier.

The '82 (12AU7) is another type that's more an RF type, and most of the circuits featuring it are RF stages (oscillator/mixers or product detectors, frequency multipliers, PP drivers, or digital and quasi-digital applications (Schmidt triggers, astable multivibrators, one-shots, flip-flops, RS latches). Though rated like a 6FQ7, they aren't as linear, though they show up often in guitar amp designs, so no price break for using them as cathode follower grid drivers there.

 

[nigel pearson]

Do you mean the ECC88 being more linear than the ECC81 - which is no surprise since ECC81 was deliberately constructed with variable grid pitch, i.e. variable mu, while ECC88 is intended to be linear - it's non-linear cousin being the ECC189. Or do you mean in combination with a GU50, used as a driver for it?

In the Alex Kitic design ( RH 34/807/6550A ) it is the ability of the ECC 81 to help the design work that is often overlooked . Michael Koster sidesteps this by using a FET in his very nice 5881 design a la Kitic + Loftin White . This got me to look favourably at pentodes . The pentode as an input device is rather nice . It can resemble all other types of device more or less . For me that nails it . It is then a matter of belief how one goes forward .

A Chinese kit amp features a Fu 50 and input pentode used as a proper pentode . Other kits are offered with triodes . Seems to me the Chinese favour the pentode . As they say it is not difficult to convert to trode .

When I converted the Kitic design to a nicer triode ( 12BH7A RCA if I remember ) it was swings and roundabouts . The feedback was reduced as was the distortion and gain . I suspect it was better although not really in the spirit of the design . An ECC 82 as cascode was roughly like a ECC 81 ( marginally better ? k2 at 95V ) . All other cascodes were less than promising . Hybrid cascode with input transistor was fine . Again much like an ECC81 . Something like the EF184 is about the best thing if wanting a Kitic amp to work better . Doubtless the whole concept is slightly wrong ? I have a very high regard for it and think the Koster version the best .

What the Kitic design says to me is . Input shunt feedback might be preferable to g2 strapping of a pentode into triode as an output device . I got told off for saying beam tetrodes are pentodes . For this argument they serve well for each other . One assumes g2 strapping is a triode . Not exactly . Thus the slightly more difficult way has some virtue . If the driver stage ( ECC81 ) isn't stopping this feedback there is a possibility of getting a free lunch . This is why the Kitic design has no cathode capacitor for the ECC 81 to keep the anode resistance high . Problem is the amp needs a gain stage . I abandoned at that point . A input pentode with shunt feedback is also a choice .

The Valve Wizard

I was very annoyed when someone got religious about me saying a shunt feedback pentode is virtually a triode . No worse , it was an ECC81 with shunt feedback starts to beat a ECC82 . I let it go . I shouldn't because someone starting out who loves music should not be told by someone that something is wrong and please return to the religion . As long as the difference is explained it is a choice . A similar debate raged for years in Wireless World that a NPN transistor would substitute a PNP . Given 100 % negaive feedback it might . Then it was noticed it might actually be better ? A small tweak suggested by Baxandale was the clincher . Over the years we have drifted backwards on that as PNP devices have improved . Now it is a choice . Slightly the all NPN design has slipped into the also ran dept . Very sad as it is in terms of being robust it is still a winner . A pentode is roughly a device like a transistor of the NPN type . PNP transistors used with valves sometimes are useful . My oscilloscope says they do very little harm and often say what is possible .

If using the spare ECC 81 of the Kitic as a gain stage try it with shunt feedback to reduce gain . It looked great on the scope . I seem to remember I got it down to only one cap in the signal path at one point doing that ?

I saw a class A 2 amp using what looked like a Kitic arrangement . Like a fool I lost it . I suspect it was a pre A2 choice ? 845 perhaps ?

 

[ilimzn]

I don't think so. The spec sheet makes no mention of this, and the plate characteristics don't look like it either. There are some triodes made for TV front ends (the 6GK5) that do mention this specifically ("gain controlled triode") and some (the 6BQ7) with plate characteristics that look suspiciously like a variable-u characteristic is an undocumented feature. Never had the occasion to use the ECC81/12AT7, but they are seen frequently as driver devices due to the higher current capability than types like the 12AX7. Decent, low distortion loadlines aren't hard to find.

Well, looking at the spec and deriving mu from the curves tells you a lot. There are also a number of different EXX81/12AT7 constructions out there (one particulairly interesting comparison is say Philips vs Tungsram). There are different ways to get variable mu aside from radically varied pitch, the ECC81 was very often in service as a IF amp where gain compression may be desirable.
As a result, yes, a decent low distortion loadline is difficult to find, but it depends on what you deem as low distortion and what swing you need. In general, you get perhaps 50-70% of the swing you can obtain from a more linear tube, but in many apps this is more than just useful. Besides, even the 6GK5 is usable for fairly linear amplification with a properly chosen loadline, the problem is it's general characteristics are far more variable across a number of them.
It all comes down what you use it for and how, there is no general principle by which you can give a categorical answer weather a triode is or is not suitable for audio.

 

[Miles Prower]

SY mentioned the Mullard 5-20. Perhaps the most elegant implementation of Mullard style circuitry is the H/K Cit. 5. For very good reason, Stu Hegeman loved high gm. The "Catch 22" here is the dwindling 12BY7 supply that is (IMO) best left to keep existing H/K Cit. 2 and Cit. 5 amps running. Fortunately, similar high gm/high bandwidth pentode types are available to the DIYer interested in building a "1 Off". The Cit. 5 can be improved by installing a 10M45S CCS in the LTP's tail.

BTW, Octal enthusiasts have the 6AC7 for something with HIGH gm.

Can't do that. With these characteristics (attached) can't say no to 12BY7As (forch, I have a bunch already and an idea for 'em).

h₂= 6.09%
h₃= 1.35%
h₄= -0.68%
THD= 6.28% (est)

Over a smaller output swing and/or use as an LTP and that distortion performance looks pretty good indeed, and it can swing a pretty good output. Expected since this was designed as a vid amp for color TVs. Only problem is that this one likes a high DC rail, which, of course, wasn't really a problem as there is no lack of high voltage in a color TV.

Another type that looks really good is the 6CB6 (attached)

h₂= 4.55%
h₃= 0
h₄= 0
THD= 4.55% (est)

Who says pents have to make higher order harmonics? As an LTP splitter, this should be pretty clean.

This one ain't so bad either (6DJ6):

h₂= 5.43%
h₃= 0.054%
h₄= 0.95%
THD= 5.5%

No need to be afraid of using pents for an audio design.

 

[nigel pearson]

Well said .

I have been experimenting with EL 84 as a signal valve . Nothing worth writing down yet . Slightly dead Mullards can have nice new lives doing this . 10 mA typical as a pentode . 320 V 10 K load . 51 R cathode resistor . 2 x 220 K to G2 is a place to start ( about 120 V DC , nice film cap ) . If anyone has any more on this please say . EF 184 ish . I have a hunch this might work in the RH34 design in place of ECC81 . The Rp of a pentode should be OK even when a power type . As a V to I converter it should have plenty to offer and be able to push . Rp should be high enough to allow via the anode feedback ( plate to plate so called ) .

 

[45]

I saw a class A 2 amp using what looked like a Kitic arrangement . Like a fool I lost it . I suspect it was a pre A2 choice ? 845 perhaps ?

Class A2 requires low source impedance. No coupling caps are allowed, power drive is required.
The more you want to push into positive grid the lower the source impedance has to be. Distortion soon becomes proportional to source impedance.

 

[45]

Well said .

I have been experimenting with EL 84 as a signal valve . Nothing worth writing down yet .

As you like, me as well, the ECL/PCL series you should know that Diego Nardi in Italy, he was the official AN tech in Italy when Kondo was still alive, makes a nice phono preamp using a PCL82 : the triode as second gain stage and the pentode as output device in a kind of hybrid anti-SRPP (loading the cathode with a PNP transistor and taking the output at the emitter). That output stage is tricky though and requires ferrite rings into the transistor base and pentode grid together with very tight packing (uses a very small board) to avoid RF oscillation. If you can do that it works nicely. The normal CF works fine as well.
Here it is:
Diego Nardi - preamplificatore Phono Hit'r schemi
Diego Nardi - preamplificatore Phono Hit'r

 

[TheGimp]

I hate to open a new thread for this so I'm posting here as it seems to be along the lines of this thread.


I just finished disecting an old phonograph I got at a flea market. Manufacturer markings are gone. It is a "Death-Trap so I decided to tear it apart.

Turntable selectable for 78, 45, 33 rpm, single channel, with AM radio.

Tube compiment is the typical AA5 type with 12BA6 RF amp, 12BE6 IF Amp, 12AV6 det/audio amp grid leak biased, 50C5 SE output, and 35W4 Rectifier.


Nothing strange here.

I tried firing it up with dead silence. Speaker open. I tacked another in and got nothing but hum.

Time to salvage the tuner cap, OPT, and whatever I could.

Strange output transformer. Three leads on the primary.

Trace circuit - UL output with a 50C5!!!!!

Has anyone ever seen this in a portable recod player/am radio or such?

Transformer has three lines with markings AT-14, 22PGD03, 465-532

 

[Miles Prower]

^
|

Can't say I have. Those kinds of things were designed to be cheap, cheap, cheap, and not for any sort of sonic excellence. The OPTs for SE 50C5s are tiny with horrid frequency response. That seemed to be a means of hum suppression: roll off the lows below 200Hz or so. (That's about the only application as the spec sheet mentions nothing of PP designs at all.) The odd thing about the specs for the 50C5 is two sets of plate characteristics: one for V₂₂= 110V (most common usage) and another for V₂₂= 90V that's a good deal more linear, though you sacrifice some output power. Like the manufacturer had better sonic performance in mind?

You can find good PP, Class A, loadlines for the type that look promising. Of course, you'd need better OPTs, and these might have to be custom jobs to use these right.

 

[DF96]

Double-check the OPT wiring. More likely than UL is a hum-cancelling arrangement.